\u00bfC\u00f3mo se puede impedir que los militares construyan por error SkyNet a partir de las pel\u00edculas Terminator? Resulta que existe una aplicaci\u00f3n para eso y el Pent\u00e1gono quiere que el p\u00fablico la use. Hace tres a\u00f1os, el Departamento de Defensa adopt\u00f3 cinco principios generales sobre c\u00f3mo podr\u00eda aplicar \u00e9ticamente la inteligencia artificial a misiones militares, desde eliminar el sesgo racial en los datos de entrenamiento de los algoritmos hasta incorporar interruptores de apagado en caso de que la IA salga mal.<\/p>\n
WASHINGTON \u2014 How do you stop the military from building SkyNet from the Terminator movies by mistake? It turns out there\u2019s an app for that \u2014 and the Pentagon wants the public to use it.<\/p>\n
Three years ago, the Department of Defense adopted\u00a0five broad principles<\/a>\u00a0for how it could ethically apply artificial intelligence to military missions, from eliminating racial bias in algorithms\u2019 training data to building in\u00a0kill switches in case AI goes awry<\/a>. Last year, the Pentagon\u2019s Chief Digital & AI Office (CDAO) turned those principles into a\u00a0detailed implementation strategy for \u201cresponsible AI\u201d<\/a>\u00a0and promised to build a toolkit for weapons acquisition officials seeking to apply the general guidance to their specific programs.\u00a0And\u00a0last night, a senior advisor to the CDAO\u2019s Responsible AI team said\u00a0that toolkit<\/a>\u00a0will be released \u201cvery soon\u201d \u2014 and it\u2019ll be available online to everyone interested in, or worried about, what the DoD is doing.<\/p>\n \u201cIt\u2019s a web app you can get to, it\u2019s publicly accessible,\u201d Matthew K. Johnson told reporters at a roundtable hosted by consulting giant Booz Allen Hamilton. \u201cIt needs to be publicly releasable and usable, so our industry partners know exactly what our expectations are [and] so the public knows exactly how the DoD is thinking.\u201d<\/p>\n \u201cIt\u2019s coming out soon,\u201d Johnson told reporters. \u201cI can\u2019t say when but it is very, very soon.\u201d \u00a0And even non-US users are welcome, he went on: \u201cA key piece of our defense strategy is integration and interoperability with partners [abroad]. It\u2019s critical for projects like\u00a0JADC2<\/a>.\u201d<\/p>\n Promoting understanding, transparency and cooperation among US officials, defense contractors and foreign allies is just part of the Pentagon\u2019s ambitious agenda here, Johnson told a public forum earlier in the evening. The grand strategic plan, he said, is to use the US military\u2019s buying power to nudge evolving technology towards American ideals of openness and privacy \u2014 and away from the Chinese Communist Party\u2019s authoritarian vision of AI as a tool for control and propaganda.<\/p>\n \u201cWe\u2019re really trying to shape this overall ecosystem because, surprise, there are others who are trying to shape this ecosystem \u2026 a lot like Belts and Roads,\u201d he said, alluding to Xi Jinping\u2019s much-touted \u201cBelt and Road<\/a>\u201d initiative to recenter global trade on China. \u201cResponsible AI looks like this kind of soft, cushy, amorphous thing, but, actually, I think it\u2019s a tremendous source of soft power \u2026 if we can spread the technology that spreads US values.\u201d<\/p>\n \u201cOne of the things our team thinks about a lot is, how do we incentivize responsible AI?\u201d Johnson said. \u201cWe\u2019ve been thinking primarily in terms of carrots rather than sticks, [and] one of the big carrots we have with DoD and a $900 billion a year budget is funding.<\/p>\n \u201c[So] how do we set in place these very clear requirements and criteria to demonstrate that your technology is aligned with the DoD AI ethical principle and our values \u2026 instead of some vague handwaving?\u201d he went on.<\/p>\n One step was last year\u2019s \u201cresponsible AI\u201d implementation plan, which managed to translate the five general principles adopted in 2020 into 64 detailed \u201clines of effort,\u201d from educating the workforce to publishing \u201cmodel cards<\/a>\u201d explaining how each AI model works. But such formal plans are still \u201cstatic documents\u201d that program managers and their staffs may struggle to apply to their unique situation.<\/p>\n So the CDAO has built an online tool to walk the user through the self-assessment process, helping them figure out how to implement the five principles and 64 lines of effort in a specific program. The software is intended to cover every stage of a program, from the initial brainstorming through development and fielding to the final retirement of a technology, and to tailor the guidance it provides to the user\u2019s specific responsibilities on the program.<\/p>\n For any given user, Johnson told reporters,\u00a0the software is meant to answer a host of questions: \u201cWhat are the responsible AI activities that I need to do? How do I do those activities? What tools do I need to use? How do I identify risks? how do I identify opportunities? [How do I] leverage all those and document those?\u201d<\/p>\n It\u2019ll take some time to work out all the bugs, Johnson acknowledged, and no one will be forced to use the tool, at least in its initial form. \u201cAt this stage, it\u2019s just a voluntary tool or a voluntary resource, so by no means is it mandated,\u201d he emphasized to reporters. \u201cWhat we are releasing is a starting point, it\u2019s a Version 1, and we are going to be validating it on a number of DoD use cases\u2026and continually updating.\u201d<\/p>\n